CN110135301A - Traffic sign recognition methods, device, equipment and computer-readable medium - Google Patents
Traffic sign recognition methods, device, equipment and computer-readable medium Download PDFInfo
- Publication number
- CN110135301A CN110135301A CN201910362985.5A CN201910362985A CN110135301A CN 110135301 A CN110135301 A CN 110135301A CN 201910362985 A CN201910362985 A CN 201910362985A CN 110135301 A CN110135301 A CN 110135301A
- Authority
- CN
- China
- Prior art keywords
- sample image
- image
- original
- traffic sign
- original sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/582—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of traffic signs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Databases & Information Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
Embodiment of the disclosure is related to traffic sign recognition methods, device, equipment and computer-readable medium.In one approach, the original image with traffic sign is obtained；Pass through machine learning model, target image is generated based on original image, machine learning model is trained based on multipair sample image, wherein every a pair of sample image includes the modification sample image after having the original sample image of traffic sign and modifying to original sample image；And traffic sign identification model is at least trained based on target image.
Description
Technical field
Embodiment of the disclosure is generally related to automatic Pilot or auxiliary drives, and relates more specifically to traffic sign identification
Method, apparatus, equipment and computer-readable medium.
Background technique
Traffic sign identification is automatic Pilot, auxiliary drives or one of core function of unmanned vehicle, be directly related to passenger and
Pedestrains safety.It is completed however, current traffic sign is identified usually using neural network.However, data set relies on open source data
Collection and artificial acquisition, data volume are extremely limited.Since traffic sign data are seriously deficient, training sample is easy to cause neural network less
Over-fitting occurs, this has seriously affected the effect of traffic sign identification.
Accordingly, it is desirable to provide a kind of model for training identification traffic sign at least partly solving above-mentioned technical problem
Scheme.
Summary of the invention
In accordance with an embodiment of the present disclosure, a kind of scheme relevant to traffic sign identification model is provided.
In the disclosure in a first aspect, providing a kind of method for training traffic sign identification model.This method comprises:
Obtain the original image with traffic sign；By machine learning model, target image, the machine are generated based on the original image
Device learning model is trained based on multipair sample image, wherein every a pair of of sample image packet in the multipair sample image
Modification sample image after including the original sample image with traffic sign and modifying to the original sample image；And
The traffic sign identification model is at least trained based on the target image.
In the second aspect of the disclosure, a kind of method of traffic sign for identification is provided.This method comprises: obtaining wait know
Other image；And the images to be recognized is identified by traffic sign identification model, the traffic sign identification model is to pass through root
The training according to method described in first aspect.
In the third aspect of the disclosure, provide a kind of for training the device of traffic sign identification model.The device includes:
Module is obtained, is configured as obtaining the original image with traffic sign；Generation module is configured as through machine learning model,
Based on the original image generate target image, the machine learning model be trained based on multipair sample image, wherein
Every a pair of of sample image in the multipair sample image includes having the original sample image of traffic sign and to the original sample
This image modify after modification sample image；And training module, be configured as at least based on the target image come
The training traffic sign identification model.
In the fourth aspect of the disclosure, a kind of device of traffic sign for identification is provided.The device includes: figure to be identified
As obtaining module, it is configured as obtaining images to be recognized；And identification module, it is configured as knowing by traffic sign identification model
The not described images to be recognized, the traffic sign identification model are the training and method according to first aspect.
At the 5th aspect of the disclosure, a kind of electronic equipment is provided.The electronic equipment includes: one or more processing
Device；And memory, for storing one or more programs, when one or more of programs are by one or more of processing
When device executes, so that the electronic equipment realizes the method according to first aspect.
At the 6th aspect of the disclosure, a kind of electronic equipment is provided.The electronic equipment includes: one or more processing
Device；And memory, for storing one or more programs, when one or more of programs are by one or more of processing
When device executes, so that the electronic equipment realizes the method according to second aspect.
At the 7th aspect of the disclosure, a kind of computer readable storage medium is provided, computer program is stored thereon with,
The method according to first aspect is realized when described program is executed by processor.
In the eighth aspect of the disclosure, a kind of computer readable storage medium is provided, computer program is stored thereon with,
The method according to second aspect is realized when described program is executed by processor.
It should be appreciated that content described in Summary be not intended to limit embodiment of the disclosure key or
Important feature, it is also non-for limiting the scope of the present disclosure.The other feature of the disclosure will become easy reason by description below
Solution.
Detailed description of the invention
It refers to the following detailed description in conjunction with the accompanying drawings, the above and other feature, advantage and aspect of each embodiment of the disclosure
It will be apparent.In the accompanying drawings, the same or similar appended drawing reference indicates the same or similar element, in which:
Fig. 1 is shown can be in the schematic diagram for the exemplary environments for wherein realizing embodiment of the disclosure；
Fig. 2A shows the process of the method for training traffic sign identification model according to some embodiments of the present disclosure
Figure；
Fig. 2 B shows the flow chart of the method for the traffic sign for identification according to some embodiments of the present disclosure；
Fig. 3 shows the schematic diagram that the training of network is generated according to the confrontation of some embodiments of the present disclosure；
Fig. 4 shows the schematic diagram that the training of network is generated according to the confrontation of some embodiments of the present disclosure；
Fig. 5 A shows the box of the device for training traffic sign identification model according to some embodiments of the present disclosure
Figure；
Fig. 5 B shows the block diagram of the device of the traffic sign for identification according to some embodiments of the present disclosure；And
Fig. 6 shows the block diagram that can implement the electronic equipment of some embodiments of the present disclosure.
Specific embodiment
Embodiment of the disclosure is more fully described below with reference to accompanying drawings.Although showing the certain of the disclosure in attached drawing
Embodiment, it should be understood that, the disclosure can be realized by various forms, and should not be construed as being limited to this
In the embodiment that illustrates, providing these embodiments on the contrary is in order to more thorough and be fully understood by the disclosure.It should be understood that
It is that being given for example only property of the accompanying drawings and embodiments effect of the disclosure is not intended to limit the protection scope of the disclosure.
As described above, at present since data set is less, the problem of causing traffic sign identification model to be easy to appear over-fitting.Separately
Outside, although there are Uniform provisions in country to the pattern specification of traffic, various regions executive condition is different, the pattern multiplicity of traffic sign,
Available data collection is difficult to exhaustive.In addition, actual environment and ideal traffic sign also will appear various deviations, thus to knowledge
Other effect causes serious influence.
Regarding to the issue above and other possible potential problems, embodiment of the disclosure provide a kind of for training friendship
The scheme of logical board identification model.In this scenario, by machine learning model, target image is generated based on original image, it is described
Machine learning model is trained based on multipair sample image.Every a pair of of sample image in the multipair sample image includes
Original sample image with traffic sign and the modification sample image after modifying to the original sample image.Target figure
As can be used for training traffic sign identification model.In this way it is possible to obtain sufficient training data to train traffic sign to know
Other model, thus the phenomenon that avoiding over-fitting.Herein, " traffic sign " is also referred to as traffic mark board, can be both sides of the road or
Sign board etc. of instruction traffic rules on person's road etc. at each position, such as left-hand rotation, right-hand rotation, straight trip, speed limit etc..
Embodiment of the disclosure is specifically described below in conjunction with Fig. 1-Fig. 4.Fig. 1 is shown can realize the disclosure wherein
Embodiment exemplary environment 100 schematic diagram.As shown in Figure 1, original sample image 102 can be the true bat of traffic sign
Image is taken the photograph, the designed image of traffic sign is also possible to.It can modify to original sample image 102, to generate modification sample
Image 104.For example, the illumination of original sample image 102, shooting angle, shooting distance, clarity can be modified and blocked
One or more aspect modifies sample image 104 to generate.In some embodiments, it can be come from by computer program
Dynamic modification original sample image 102, for example, can be modified by means of OpenCV.It alternatively, can also be with manual modification original
Beginning sample image 102, to increase flexibility.
Original sample image 102 and modification sample image 104 can be supplied to machine learning model 106, to carry out to it
Training.Machine learning model 106 after training could be aware that from original sample image 102 to modification sample image 104
Mapping relations.In this way, the original image of traffic sign 110 is supplied to machine learning model 106.Machine learning model
106 obtain modification image corresponding with original image 110, also referred to as target image.Target image is supplied to identification model 108,
Identification model 108 is trained.Identification model 108 can be more disaggregated models, identify to received image, with true
Determine the type of traffic sign.Original image 110 can also be supplied to identification model 108 together with target image, to identification model 108
It is trained.
Fig. 2A shows the stream of the method 200 for training traffic sign identification model according to some embodiments of the present disclosure
Cheng Tu.Method 200 is described below with reference to the exemplary environments 100 of Fig. 1, it being understood, however, that, method 200 can also answer
For any other suitable environment except exemplary environments 100 as shown in Figure 1.
In frame 202, the original image 110 with traffic sign is obtained.Original image can be by various methods to obtain
, for example, the image comprising traffic sign around the road shot in vehicle travel process, for example, a frame of video.
Target image is generated based on original image 110 by machine learning model 106 in frame 204.Machine learning mould
Type 106 is trained based on multipair sample image, and every a pair of sample image includes the original sample image 102 with traffic sign
With modify to original sample image 102 after modification sample image 104.By the training to machine learning model 106,
Machine learning model 106 can be indicated from original sample image 102 to the mapping relations modification sample image 104.Pass through
This mapping relations, which is applied to original image 110, can obtain target image.It can be in one or more aspects to original sample
This image 102 is modified to obtain and modify sample image 104.It will be described in greater detail below according to several embodiments.
In one embodiment, modification sample image 104 be modify to the shooting angle of original sample image 102 and
It obtains.For example, can be by the mobile certain angle of the shooting angle of original sample image 102, to be reflected in various different angles
Lower shooting traffic sign image obtained therein.
In one embodiment, modification sample image 104 is to modify and obtain to the illumination of original sample image 102
's.For example, original sample image 102 can be and shoot under the conditions of illumination is preferable, so as to by original sample image
102 illumination condition is revised as poor illumination condition, modifies sample image 104 to obtain.
In one embodiment, modification sample image 104 be modify to the shooting distance of original sample image 102 and
It obtains.For example, the shooting distance of original sample image 102 may be relatively close, however, it is expected that being known in farther away situation
Not, so that automatic driving vehicle can provide in the biggish situation of better judgement, especially Vehicle Speed.
In one embodiment, modification sample image 104 is to modify and obtain to the clarity of original sample image 102
?.For example, original sample image 102 may be the higher image of clarity, it can be by the clarity of original sample image 102
It is reduced, to obtain modification sample image 104.
In one embodiment, modification sample image 104 is to apply to block and obtain at random to original sample image 102
's.For example, occlusion effect can be simulated by the way that some pixels in original sample image 102 are adjusted to black.Traffic sign
Various circumstance of occlusion are likely to occur, for example, small advertisement, rain erosion etc..Modification sample image 104 can simulate these and block feelings
Condition, to increase the diversity of sample.
Several embodiments in the modification original sample image 102 of one or more aspects are described above.It should be appreciated that
These embodiments are simultaneously non-exclusive, but can be combined each other, to generate further new embodiment.
In some embodiments, machine learning model 106 can be confrontation and generate network, for example, depth convolution is to antibiosis
At network.It includes generator and arbiter that confrontation, which generates network, and generator can be based on depth deconvolution network, and arbiter can
To be based on depth convolutional network.Confrontation generate network embodiment in, can by generator 304 from original image 110 come
Generate target image.
It is generated in network in confrontation, the sample that generator generates is known as dummy copy, authentic specimen is known as true sample.Sentence
The sample that other device is responsible for judgement input is true sample or dummy copy.The training objective of generator is that the sample generated is increasingly forced
Very, arbiter is cheated until that can mix the spurious with the genuine.The training objective of arbiter be can correctly distinguish as far as possible it is true and false.Confrontation generates
It is after network training as a result, generator can generate it is a large amount of extremely close to the samples of modification sample images, it might even be possible to cheat
Cross arbiter.
The embodiment that confrontation generates network is introduced below with reference to Fig. 3-Fig. 4, wherein Fig. 3 shows the feelings of true sample
The case where condition, Fig. 4 shows dummy copy.
As shown in figure 3, arbiter 302 receives original sample image 102 and modification sample image 104, and determine the two
Between whether correspond to.In the case where this true sample, the training objective of arbiter 302 is that there are corresponding relationships for both judgements.
As shown in figure 4, generator 304 generates false modification image 306 based on original sample image 102.Arbiter 302 connects
False modification sample image 306 and original sample image 102 are received, and whether determination corresponds between the two.In the feelings of this dummy copy
Under condition, the training objective of arbiter 302 is that there is no corresponding relationships for both judgements, and the training objective of generator 304 is to generate
Making both 302 false judgments of arbiter as far as possible, there are corresponding relationships.
In frame 206, traffic sign identification model 108 is at least trained based on target image.For example, target image can be added
Enter into the training set of traffic sign identification model 108, it is come together to train traffic sign identification model 108 with other images.It hands over
Logical board identification model 108 can be neural network model, for example, deep neural network model.Can be used it is being currently known or
Any suitable method of person's exploitation in the future trains traffic sign identification model 108, for example, stochastic gradient descent method (SGD) etc..
Fig. 2 B shows the flow chart of the method 250 of the traffic sign for identification according to some embodiments of the present disclosure.With
Lower exemplary environments 100 by conjunction with Fig. 1 describe method 250, it being understood, however, that, method 250 also can be applied to such as figure
Any other suitable environment except exemplary environments 100 shown in 1.
In frame 252, images to be recognized is obtained.Images to be recognized can be the image that automatic driving vehicle acquires in real time or
Video frame.Images to be recognized can be locally processed in vehicle, or can also be transmitted to cloud, to be handled beyond the clouds.
In frame 254, images to be recognized is identified by traffic sign identification model 108.Traffic sign identification model 108 can be
It is trained according to method 200 as shown in Figure 2 A.In this way it is possible to determine the class of the traffic sign in images to be recognized
Type.
Embodiment of the disclosure can effectively overcome traffic sign data seriously deficient, and training sample leads to network over-fitting less
The shortcomings that, data sample is effectively enhanced, and make the generalization ability of traffic sign identification network and robustness stronger, overcome
Different angle, different illumination, different distance, different clarity and rain erosion, small advertisement etc. cover the influence to identification, mention
High traffic sign discrimination, practical value with higher.
Fig. 5 A shows the side of the device 500 for training traffic sign identification model according to some embodiments of the present disclosure
Block diagram.Device 500 can be used to realize method 200 as shown in Figure 2 A.
Device 500 includes obtaining module 502, is configured as obtaining the original image with traffic sign.
Device 500 includes generation module 504, is configured as generating by machine learning model based on the original image
Target image, the machine learning model is trained based on multipair sample image, wherein in the multipair sample image
Every a pair of sample image includes after having the original sample image of traffic sign and modifying to the original sample image
Modify sample image.
In some embodiments, at least one of below executing, to obtain and the pairs of modification sample of the original sample image
This image: the shooting angle of the original sample image is modified；Modify the illumination of the original sample image；It modifies described original
The shooting distance of sample image；Modify the clarity of the original sample image；And the original sample image is applied at random
Add and blocks.
In some embodiments, the machine learning model includes that confrontation generates network.
In some embodiments, it includes generator and arbiter that the confrontation, which generates network, and the confrontation generates net
Network is trained as follows: false modification image is generated based on the original sample image by the generator；And pass through by
The multipair sample image is as true sample and using the pairs of original sample image and the false modification image as false sample
Trained the confrontation to generate network originally.
In some embodiments, generation module 204 includes: maker module, is configured as through the generator from institute
Original image is stated to generate the target image.
Device 500 includes training module 506, is configured as that the traffic sign is at least trained to know based on the target image
Other model.
Fig. 5 B shows the block diagram of the device 550 of the traffic sign for identification according to some embodiments of the present disclosure.Dress
Setting 550 can be used to realize method 250 as shown in Figure 2 B.
As shown in Figure 5 B, device 550 includes that images to be recognized obtains module 552, is configured as obtaining images to be recognized.
Device 550 further includes identification module 554, is configured as identifying the figure to be identified by traffic sign identification model
Picture, traffic sign identification model are by the training of method 200 as shown in Figure 2 A.
Fig. 6 shows the schematic block diagram that can be used to implement the equipment 600 of embodiment of the disclosure.Such as Fig. 1 institute
The exemplary environments 100 and device 500 as shown in Figure 5 shown can be realized by equipment 600.As shown in fig. 6, equipment 600
It, can be according to the computer program instructions being stored in read-only memory (ROM) 602 including central processing unit (CPU) 601
Or the computer program instructions in random access storage device (RAM) 603 are loaded into from storage unit 608, it is various suitable to execute
When movement and processing.In RAM 603, it can also store equipment 600 and operate required various programs and data.CPU 601,
ROM 602 and RAM 603 is connected with each other by bus 604.Input/output (I/O) interface 605 is also connected to bus 604.
Multiple components in equipment 600 are connected to I/O interface 605, comprising: input unit 606, such as keyboard, mouse etc.；
Output unit 607, such as various types of displays, loudspeaker etc.；Storage unit 608, such as disk, CD etc.；And it is logical
Believe unit 609, such as network interface card, modem, wireless communication transceiver etc..Communication unit 609 allows equipment 600 by such as
The computer network of internet and/or various telecommunication networks exchange information/data with other equipment.
Each process as described above and processing, such as method 200 can be executed by processing unit 601.For example, one
In a little embodiments, method 200 can be implemented as computer software programs, be tangibly embodied in machine readable media, such as
Storage unit 608.In some embodiments, some or all of of computer program can be via ROM 602 and/or communication unit
Member 609 and be loaded into and/or be installed in equipment 600.When computer program is loaded into RAM 603 and is executed by CPU 601
When, the one or more steps of method as described above 200 can be executed.Alternatively, in other embodiments, CPU 601 can
By by other it is any it is appropriate in a manner of (for example, by means of firmware) be configured as execution method 200.
The disclosure can be method, equipment, system and/or computer program product.Computer program product may include
Computer readable storage medium, containing the computer-readable program instructions for executing various aspects of the disclosure.
Computer readable storage medium, which can be, can keep and store the tangible of the instruction used by instruction execution equipment
Equipment.Computer readable storage medium for example can be-- but it is not limited to-- storage device electric, magnetic storage apparatus, optical storage
Equipment, electric magnetic storage apparatus, semiconductor memory apparatus or above-mentioned any appropriate combination.Computer readable storage medium
More specific example (non exhaustive list) includes: portable computer diskette, hard disk, random access memory (RAM), read-only deposits
It is reservoir (ROM), erasable programmable read only memory (EPROM or flash memory), static random access memory (SRAM), portable
Compact disk read-only memory (CD-ROM), digital versatile disc (DVD), memory stick, floppy disk, mechanical coding equipment, for example thereon
It is stored with punch card or groove internal projection structure and the above-mentioned any appropriate combination of instruction.Calculating used herein above
Machine readable storage medium storing program for executing is not interpreted that instantaneous signal itself, the electromagnetic wave of such as radio wave or other Free propagations lead to
It crosses the electromagnetic wave (for example, the light pulse for passing through fiber optic cables) of waveguide or the propagation of other transmission mediums or is transmitted by electric wire
Electric signal.
Computer-readable program instructions as described herein can be downloaded to from computer readable storage medium it is each calculate/
Processing equipment, or outer computer or outer is downloaded to by network, such as internet, local area network, wide area network and/or wireless network
Portion stores equipment.Network may include copper transmission cable, optical fiber transmission, wireless transmission, router, firewall, interchanger, gateway
Computer and/or Edge Server.Adapter or network interface in each calculating/processing equipment are received from network to be counted
Calculation machine readable program instructions, and the computer-readable program instructions are forwarded, for the meter being stored in each calculating/processing equipment
In calculation machine readable storage medium storing program for executing.
Computer program instructions for executing disclosure operation can be assembly instruction, instruction set architecture (ISA) instructs,
Machine instruction, machine-dependent instructions, microcode, firmware instructions, condition setup data or with one or more programming languages
The source code or object code that any combination is write, the programming language include the programming language-of object-oriented such as
Smalltalk, C++ etc., and conventional procedural programming languages-such as " C " language or similar programming language.Computer
Readable program instructions can be executed fully on the user computer, partly execute on the user computer, be only as one
Vertical software package executes, part executes on the remote computer or completely in remote computer on the user computer for part
Or it is executed on server.In situations involving remote computers, remote computer can pass through network-packet of any kind
It includes local area network (LAN) or wide area network (WAN)-is connected to subscriber computer, or, it may be connected to outer computer (such as benefit
It is connected with ISP by internet).In some embodiments, by utilizing computer-readable program instructions
Status information carry out personalized customization electronic circuit, such as programmable logic circuit, field programmable gate array (FPGA) or can
Programmed logic array (PLA) (PLA), the electronic circuit can execute computer-readable program instructions, to realize each side of the disclosure
Face.
Referring herein to according to the flow chart of the method, apparatus (system) of the embodiment of the present disclosure and computer program product and/
Or block diagram describes various aspects of the disclosure.It should be appreciated that flowchart and or block diagram each box and flow chart and/
Or in block diagram each box combination, can be realized by computer-readable program instructions.
These computer-readable program instructions can be supplied to general purpose computer, special purpose computer or other programmable datas
The processing unit of processing unit, so that a kind of machine is produced, so that these instructions are passing through computer or other programmable numbers
When being executed according to the processing unit of processing unit, produces and provided in one or more boxes in implementation flow chart and/or block diagram
Function action device.These computer-readable program instructions can also be stored in a computer-readable storage medium, this
A little instructions so that computer, programmable data processing unit and/or other equipment work in a specific way, thus, be stored with finger
The computer-readable medium of order then includes a manufacture comprising the one or more side in implementation flow chart and/or block diagram
The instruction of the various aspects of function action specified in frame.
Computer-readable program instructions can also be loaded into computer, other programmable data processing units or other
In equipment, so that series of operation steps are executed in computer, other programmable data processing units or other equipment, to produce
Raw computer implemented process, so that executed in computer, other programmable data processing units or other equipment
Instruct function action specified in one or more boxes in implementation flow chart and/or block diagram.
The flow chart and block diagram in the drawings show system, method and the computer journeys according to multiple embodiments of the disclosure
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
One module of table, program segment or a part of instruction, the module, program segment or a part of instruction include one or more use
The executable instruction of the logic function as defined in realizing.In some implementations as replacements, function marked in the box
It can occur in a different order than that indicated in the drawings.For example, two continuous boxes can actually be held substantially in parallel
Row, they can also be executed in the opposite order sometimes, and this depends on the function involved.It is also noted that block diagram and/or
The combination of each box in flow chart and the box in block diagram and or flow chart, can the function as defined in executing or dynamic
The dedicated hardware based system made is realized, or can be realized using a combination of dedicated hardware and computer instructions.
The presently disclosed embodiments is described above, above description is exemplary, and non-exclusive, and
It is not limited to the disclosed embodiment.Without departing from the scope and spirit of illustrated each embodiment, for this skill
Many modifications and changes are obvious for the those of ordinary skill in art field.The selection of term used herein, purport
In the principle, practical application or improvement to the technology in market for best explaining each embodiment, or make the art
Other those of ordinary skill can understand various embodiments disclosed herein.
Claims (18)
1. a kind of method for training traffic sign identification model, comprising:
Obtain the original image with traffic sign；
By machine learning model, target image is generated based on the original image, the machine learning model is based on multipair
Sample image is trained, wherein every a pair of of sample image in the multipair sample image includes the original sample with traffic sign
This image and the modification sample image after modifying to the original sample image；And
The traffic sign identification model is at least trained based on the target image.
2. according to the method described in claim 1, wherein execute it is at least one of following, with obtain with the original sample image at
Pair modification sample image:
Modify the shooting angle of the original sample image；
Modify the illumination of the original sample image；
Modify the shooting distance of the original sample image；
Modify the clarity of the original sample image；And
The original sample image is applied at random and is blocked.
3. according to the method described in claim 1, wherein the machine learning model includes that confrontation generates network.
4. according to the method described in claim 3, it includes generator and arbiter that wherein the confrontation, which generates network, and described
It is trained as follows that confrontation, which generates network:
False modification image is generated based on the original sample image by the generator；And
By scheming using the multipair sample image as true sample and by the pairs of original sample image and the false modification
As training the confrontation to generate network as dummy copy.
5. according to the method described in claim 4, wherein generating the target image and including:
The target image is generated from the original image by the generator.
6. method according to any one of claims 1-5, wherein the machine learning model is indicated from the original sample
This image arrives the mapping relations between the modification sample image,
Wherein generating the target image includes: to obtain the mesh by the way that the mapping relations are applied to the original image
Logo image.
7. a kind of method of traffic sign for identification, comprising:
Obtain images to be recognized；And
Identify that the images to be recognized, the traffic sign identification model are by wanting according to right by traffic sign identification model
Seek method described in any one of 1-6 and training.
8. a kind of for training the device of traffic sign identification model, comprising:
Module is obtained, is configured as obtaining the original image with traffic sign；
Generation module, is configured as through machine learning model, generates target image, the engineering based on the original image
Practising model is trained based on multipair sample image, wherein every a pair of of sample image in the multipair sample image includes tool
Modification sample image after having the original sample image of traffic sign and modifying to the original sample image；And
Training module is configured as at least training the traffic sign identification model based on the target image.
9. device according to claim 8, wherein execute it is at least one of following, with obtain with the original sample image at
Pair modification sample image:
Modify the shooting angle of the original sample image；
Modify the illumination of the original sample image；
Modify the shooting distance of the original sample image；
Modify the clarity of the original sample image；And
The original sample image is applied at random and is blocked.
10. device according to claim 8, wherein the machine learning model includes that confrontation generates network.
11. device according to claim 10, wherein it includes generator and arbiter that the confrontation, which generates network, and institute
It is trained as follows for stating confrontation and generating network:
False modification image is generated based on the original sample image by the generator；And
By scheming using the multipair sample image as true sample and by the pairs of original sample image and the false modification
As training the confrontation to generate network as dummy copy.
12. device according to claim 11, wherein the generation module includes:
Maker module is configured as generating the target image from the original image by the generator.
13. the device according to any one of claim 8-12, wherein the machine learning model is indicated from described original
Sample image to it is described modification sample image between mapping relations,
Wherein the generation module includes generating submodule, is configured as by the way that the mapping relations are applied to the original graph
As obtaining the target image.
14. a kind of device of traffic sign for identification, comprising:
Images to be recognized obtains module, is configured as obtaining images to be recognized；And
Identification module is configured as identifying the images to be recognized by traffic sign identification model, and the traffic sign identifies mould
Type is by method training according to claim 1 to 6.
15. a kind of electronic equipment, the electronic equipment include:
One or more processors；And
Memory, for storing one or more programs, when one or more of programs are by one or more of processors
When execution, so that the electronic equipment realizes method according to claim 1 to 6.
16. a kind of electronic equipment, the electronic equipment include:
One or more processors；And
Memory, for storing one or more programs, when one or more of programs are by one or more of processors
When execution, so that the electronic equipment is realized according to the method for claim 7.
17. a kind of computer readable storage medium is stored thereon with computer program, realization when described program is executed by processor
Method according to claim 1 to 6.
18. a kind of computer readable storage medium is stored thereon with computer program, realization when described program is executed by processor
According to the method for claim 7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910362985.5A CN110135301B (en) | 2019-04-30 | 2019-04-30 | Traffic sign recognition method, device, equipment and computer readable medium |
| US16/811,121 US20200349369A1 (en) | 2019-04-30 | 2020-03-06 | Method and apparatus for training traffic sign idenfication model, and method and apparatus for identifying traffic sign |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910362985.5A CN110135301B (en) | 2019-04-30 | 2019-04-30 | Traffic sign recognition method, device, equipment and computer readable medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110135301A true CN110135301A (en) | 2019-08-16 |
| CN110135301B CN110135301B (en) | 2022-02-22 |
Family
ID=67575964
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910362985.5A Active CN110135301B (en) | 2019-04-30 | 2019-04-30 | Traffic sign recognition method, device, equipment and computer readable medium |
Country Status (2)
| Country | Link |
|---|---|
| US (1) | US20200349369A1 (en) |
| CN (1) | CN110135301B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112418390A (en) * | 2019-08-23 | 2021-02-26 | 国际商业机器公司 | Comparative interpretation of images using monotonic attribute function |
| CN113177497A (en) * | 2021-05-10 | 2021-07-27 | 百度在线网络技术（北京）有限公司 | Visual model training method, vehicle identification method and device |
| US11605232B2 (en) | 2020-08-27 | 2023-03-14 | Robert Bosch Gmbh | System and method for road sign ground truth construction with a knowledge graph and machine learning |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116188919B (en) * | 2023-04-25 | 2023-07-14 | 之江实验室 | Test method and device, readable storage medium and electronic equipment |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106952239A (en) * | 2017-03-28 | 2017-07-14 | 厦门幻世网络科技有限公司 | image generating method and device |
| CN107154023A (en) * | 2017-05-17 | 2017-09-12 | 电子科技大学 | Face super-resolution reconstruction method based on generation confrontation network and sub-pix convolution |
| US20170323452A1 (en) * | 2014-11-18 | 2017-11-09 | Agency For Science, Technology And Research | Method and device for traffic sign recognition |
| CN108491858A (en) * | 2018-02-11 | 2018-09-04 | 南京邮电大学 | Method for detecting fatigue driving based on convolutional neural networks and system |
| CN108846355A (en) * | 2018-06-11 | 2018-11-20 | 腾讯科技（深圳）有限公司 | Image processing method, face identification method, device and computer equipment |
| CN108875511A (en) * | 2017-12-01 | 2018-11-23 | 北京迈格威科技有限公司 | Method, apparatus, system and the computer storage medium that image generates |
| CN108875766A (en) * | 2017-11-29 | 2018-11-23 | 北京旷视科技有限公司 | Method, apparatus, system and the computer storage medium of image procossing |
| CN108921089A (en) * | 2018-06-29 | 2018-11-30 | 驭势科技（北京）有限公司 | Method for detecting lane lines, device and system and storage medium |
| CN108932534A (en) * | 2018-07-15 | 2018-12-04 | 瞿文政 | A kind of Picture Generation Method generating confrontation network based on depth convolution |
| CN109241830A (en) * | 2018-07-26 | 2019-01-18 | 合肥工业大学 | It listens to the teacher method for detecting abnormality in the classroom for generating confrontation network based on illumination |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017173100A1 (en) * | 2016-03-30 | 2017-10-05 | Convenant Eyes, Inc. | Applications, systems and methods to monitor, filter and/or alter output of a computing device |
| CN111527493A (en) * | 2017-10-31 | 2020-08-11 | 利惠商业有限公司 | Using neural networks in creating apparel designs |
-
2019
- 2019-04-30 CN CN201910362985.5A patent/CN110135301B/en active Active
-
2020
- 2020-03-06 US US16/811,121 patent/US20200349369A1/en not_active Abandoned
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170323452A1 (en) * | 2014-11-18 | 2017-11-09 | Agency For Science, Technology And Research | Method and device for traffic sign recognition |
| CN106952239A (en) * | 2017-03-28 | 2017-07-14 | 厦门幻世网络科技有限公司 | image generating method and device |
| CN107154023A (en) * | 2017-05-17 | 2017-09-12 | 电子科技大学 | Face super-resolution reconstruction method based on generation confrontation network and sub-pix convolution |
| CN108875766A (en) * | 2017-11-29 | 2018-11-23 | 北京旷视科技有限公司 | Method, apparatus, system and the computer storage medium of image procossing |
| CN108875511A (en) * | 2017-12-01 | 2018-11-23 | 北京迈格威科技有限公司 | Method, apparatus, system and the computer storage medium that image generates |
| CN108491858A (en) * | 2018-02-11 | 2018-09-04 | 南京邮电大学 | Method for detecting fatigue driving based on convolutional neural networks and system |
| CN108846355A (en) * | 2018-06-11 | 2018-11-20 | 腾讯科技（深圳）有限公司 | Image processing method, face identification method, device and computer equipment |
| CN108921089A (en) * | 2018-06-29 | 2018-11-30 | 驭势科技（北京）有限公司 | Method for detecting lane lines, device and system and storage medium |
| CN108932534A (en) * | 2018-07-15 | 2018-12-04 | 瞿文政 | A kind of Picture Generation Method generating confrontation network based on depth convolution |
| CN109241830A (en) * | 2018-07-26 | 2019-01-18 | 合肥工业大学 | It listens to the teacher method for detecting abnormality in the classroom for generating confrontation network based on illumination |
Non-Patent Citations (1)
| Title |
|---|
| 王娇娇: "基于显著性与卷积神经网络的交通标志检测与识别研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112418390A (en) * | 2019-08-23 | 2021-02-26 | 国际商业机器公司 | Comparative interpretation of images using monotonic attribute function |
| CN112418390B (en) * | 2019-08-23 | 2024-06-11 | 国际商业机器公司 | Contrast interpretation of images using monotonic attribute functions |
| US11605232B2 (en) | 2020-08-27 | 2023-03-14 | Robert Bosch Gmbh | System and method for road sign ground truth construction with a knowledge graph and machine learning |
| CN113177497A (en) * | 2021-05-10 | 2021-07-27 | 百度在线网络技术（北京）有限公司 | Visual model training method, vehicle identification method and device |
| CN113177497B (en) * | 2021-05-10 | 2024-04-12 | 百度在线网络技术（北京）有限公司 | Training method of visual model, vehicle identification method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110135301B (en) | 2022-02-22 |
| US20200349369A1 (en) | 2020-11-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110135301A (en) | Traffic sign recognition methods, device, equipment and computer-readable medium | |
| Chen et al. | Touchdown: Natural language navigation and spatial reasoning in visual street environments | |
| US11135513B2 (en) | Virtual vehicle control method, model training method, control device and storage medium | |
| CN108229468A (en) | Vehicle appearance feature recognition and vehicle retrieval method, apparatus, storage medium, electronic equipment | |
| CN109658418A (en) | Learning method, device and the electronic equipment of scene structure | |
| CN109255356A (en) | A kind of character recognition method, device and computer readable storage medium | |
| CN110532925B (en) | Driver fatigue detection method based on space-time graph convolutional network | |
| JP2022531639A (en) | How to embed information in video, computer equipment and computer programs | |
| CN109684803A (en) | Man-machine verification method based on gesture sliding | |
| WO2019053052A1 (en) | A method for (re-)training a machine learning component | |
| CN108763915A (en) | Identifying code is established to generate model and generate the method, apparatus of identifying code | |
| CN109889849A (en) | Video generation method, device, medium and equipment | |
| CN109426776A (en) | Object detection based on deep neural network | |
| CN104735435A (en) | Image processing method and electronic device | |
| CN113642621A (en) | Zero sample image classification method based on generation countermeasure network | |
| CN115393599A (en) | Method, device, electronic equipment and medium for constructing image semantic segmentation model and image processing | |
| CN110349138A (en) | The detection method and device of the target object of Case-based Reasoning segmentation framework | |
| CN110399690B (en) | Subway station pedestrian simulation method and device, electronic equipment and storage medium | |
| Grau et al. | A variational deep synthesis approach for perception validation | |
| Hagn et al. | Optimized data synthesis for DNN training and validation by sensor artifact simulation | |
| CN116778455A (en) | Real-time high-precision traffic sign small target detection method based on improved YOLOv7-tiny | |
| CN113628349B (en) | AR navigation method, device and readable storage medium based on scene content adaptation | |
| CN114219804B (en) | Small sample tooth detection method based on prototype segmentation network and storage medium | |
| CN109598206A (en) | Dynamic gesture identification method and device | |
| CN116983660A (en) | Image processing method and related device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |